Low FVC/TLC in Preserved Ratio Impaired Spirometry (PRISm) is associated with features of and progression to obstructive lung disease

One quarter of individuals with Preserved Ratio Impaired Spirometry (PRISm) will develop airflow obstruction, but there are no established methods to identify these individuals. We examined the utility of FVC/TLC in identifying features of obstructive lung disease. The ratio of post-bronchodilator FVC and TLCCT from chest CT (FVC/TLCCT) among current and former smokers with PRISm (FEV1/FVC ≥ 0.7 and FEV1 < 80%) in COPDGene was used to stratify subjects into quartiles: very high, high, low, and very low. We examined the associations between FVC/TLCCT quartiles and (1) baseline characteristics, (2) respiratory exacerbations, (3) progression to COPD at 5 years, and (4) all-cause mortality. Among participants with PRISm at baseline (n = 1,131), the very low FVC/TLCCT quartile was associated with increased gas trapping and emphysema, and higher rates of progression to COPD at 5 years (36% versus 17%; p < 0.001) relative to the very high quartile. The very low FVC/TLCCT quartile was associated with increased total (IRR = 1.65; 95% CI [1.07–2.54]) and severe (IRR = 2.24; 95% CI [1.29–3.89]) respiratory exacerbations. Mortality was lower in the very high FVC/TLCCT quartile relative to the other quartiles combined. Reduced FVC/TLCCT ratio in PRISm is associated with increased symptoms, radiographic emphysema and gas trapping, exacerbations, and progression to COPD.

In PRISm, total lung capacity (TLC) may help distinguish a restrictive from an obstructive ventilatory defect, according to the American Thoracic Society-European Respiratory Society (ATS-ERS) 2005 guidelines 8 . However, a true "restrictive disease" is very unlikely in individuals with risk factors for obstruction lung disease, no interstitial lung disease and unremarkable body mass index (BMI). In COPDGene, the prevalence of PRISm is 12% despite the fact that participants with interstitial lung disease were excluded and body mass index (BMI) in PRISm individuals was slightly higher than the BMI in smokers with normal lung function 1,2 . In addition, a single center study showed that among individuals with PRISm and TLC above the lower limit of normal (LLN), only 26% had a clinical diagnosis of obstructive lung disease 9 . Moreover, only 15% of those with PRISm and TLC > LLN develop obstructive spirometry over a median follow-up time of 3 years 10 . Currently, there is no available diagnostic test in clinical practice to identify which patients with PRISm may have features classically associated with obstructive lung disease.
In obstructive lung diseases, residual volume (RV) may increase at the expense of FVC with total lung capacity (TLC) remaining normal 11 . Conversely, RV may increase with a preserved FVC resulting in an increased TLC. Both processes result in reduced FVC/TLC ratio which may antedate the development of obstruction diagnosed using standard FEV 1 /FVC criteria. A disproportionate decrease of FVC relative to TLC may occur in patients with a restrictive ventilatory defect that coexists with obstructive lung disease 12 . Thus, FVC/TLC represents a composite measure that may be able to identify an occult obstructive ventilatory defect. FVC can be readily obtained from spirometry, while TLC, typically assessed by plethysmography, can also be quantified using an inspiratory chest CT (TLC CT ), with prior studies demonstrating strong correlations with the plethysmography results 13,14 .
We hypothesized that reduced FVC/TLC ratio in PRISm is associated with clinical, functional and radiographic features of obstructive lung disease, acute respiratory events and increased mortality, and progression to COPD. To investigate our hypothesis, we examined current and former smokers with PRISm enrolled in the COPDGene study.

Methods
Data collection. We analyzed data from participants in the COPDGene study, an ongoing study conducted at multiple clinical centers through the United States (http://www.copdgene.org/). Participants were current and former smokers with ≥10 pack-years of smoking who self-identified as non-Hispanic whites (NHW) or African Americans (AA) and were between the ages of 45-80 years at enrollment. The institutional review boards at each participating center approved the study protocol and written informed consent was obtained from all participants. Details of the study protocol have been published previously 15 . Briefly, participants completed a modified American Thoracic Society Respiratory Epidemiology questionnaire, St. George's Respiratory Questionnaire (SGRQ), and 6-minute walk test (6-MWT) at the enrollment visit. Dyspnea was assessed using the modified Medical Research Council (mMRC) scale. Participants performed pre-and post-bronchodilator spirometry. The complete study protocols were performed in accordance with the relevant guidelines and regulations of American Thoracic Society-European Respiratory Society (ATS-ERS) 16 . Volumetric chest CT scans were obtained at TLC CT (maximal inspiration) and at functional residual capacity (FRC CT ) (end-tidal expiration) using multidetector CT scanners 15 . FRC and TLC% predicted were calculated based on the predicted values 17 . Percent emphysema and gas trapping were quantified using 3D Slicer software (www.airwayinspector.org) 15 .
We included participants with PRISm at enrollment. We excluded individuals with significant interstitial lung disease or bronchiectasis on chest CT, those with missing post-bronchodilator spirometry or TLC CT measurements at baseline, and participants with post-bronchodilator FVC > TLC CT at enrollment. Approximately 5 years after the enrollment visit, participants were invited for a follow-up visit that included a repeat spirometry and chest CT. Respiratory exacerbation data were collected prospectively after enrollment. Participants were contacted every 6 months after enrollment and completed a standardized questionnaire regarding respiratory exacerbations through the Longitudinal Follow Up program. Vital status was also ascertained using information from the social security death index and the Longitudinal Follow Up program.
Definitions and outcomes. PRISm was defined as post-bronchodilator FEV 1 < 80% predicted and FEV 1 / FVC ≥ 0.7. COPD was defined as post-bronchodilator FEV 1 /FVC < 0.7. The FVC/TLC CT ratio at enrollment was calculated using post-bronchodilator FVC (in liters) from spirometry, while TLC CT was measured from volumetric inspiratory chest CT scans.
Co-morbidities and medication usage were self-reported. Percent emphysema was defined by using the percentage of lung volume at TLC CT with attenuation less than −950 Hounsfield units (HU) 15 . Gas trapping was quantified as the percentage of lung volume at FRC with attenuation values less than −856 HU 15 . Parametric response mapping analysis was performed on paired registered inspiratory and expiratory images to distinguish functional small airways disease (PRM fSAD ) from emphysema by Imbio LLC (Minneapolis, MN) using lung density analysis software 18 . As previous described 19 , we defined PRM fSAD as the percentage of lung with evidence of gas trapping not due to emphysema (i.e. areas of lung with attenuation < −856 HU on expiration minus area of lung with attenuation < −950 HU on inspiration).
Change in FEV 1 between enrollment and 5-year follow up visit was calculated using post-bronchodilator spirometry. Exacerbations were defined as episodes of worsening respiratory symptoms requiring use of antibiotics and/or systemic steroids. Severe exacerbations were defined as those requiring hospitalizations or emergency room visits. Other variable definitions have been previously described 15 . Statistical analysis. We stratified PRISm participants at the enrollment visit into quartiles by FVC/TLC CT : very high, high, low, and very low. We compared the characteristics of PRISm individuals at the enrollment visit, rates of progression to COPD at the 5-year follow-up visit, and exacerbations over the time between the FVC/TLC CT quartiles. We used Spearman's rank correlation to examine changes in continuous variables with increasing FVC/TLC CT . We used the Cochran Armitage trend test to examine proportion changes with increasing FVC/TLC CT quartile.
We created multivariable logistic and linear regression models with chronic bronchitis, mMRC and SGRQ scores, radiographic measures and 6-MWT distance at the enrollment visit as the dependent variable (outcome) and post-bronchodilator FVC/TLC CT quartile at the enrollment as the independent variable (predictor). All models included the following covariates: age and current smoking status at enrollment, gender, race, pack-years smoked, body mass index (BMI), history of asthma and congestive heart failure. There were no missing values in any of the covariates. We also performed a multivariable linear and logistic regression analysis with change in FEV 1 , 6-MWT distance, radiographic measurements, and progression to COPD at the follow-up visit as the dependent variable (outcome) and post-bronchodilator FVC/TLC CT quartile at enrollment as the independent variable (predictor). We included the following covariates in these models: age and current smoking status at enrollment, gender, race, pack-years smoked, body mass index (BMI), history of asthma and congestive heart failure, and change of BMI between enrollment and follow-up visit.
For the exacerbation analysis, we created zero-inflated negative binomial models which included adjustment for age and current smoking status at enrollment, gender, race, pack-years smoked, BMI, chronic bronchitis, history of asthma and congestive heart failure. There were no missing values in any of the covariates. Follow-up time was included as an offset in the models as previously described 20 .
We used Cox proportional hazard regression models to examine the association between post-bronchodilator FVC/TLC CT quartile with all-cause mortality. Models included the following covariates: age, gender, race, smoking status, smoking pack-years, BMI, diabetes, history of asthma and congestive heart failure. There were no missing values in any of the covariates.
Ethics approval. The institutional review boards at each participating center approved the study protocol.
Details of the study protocol have been published previously 16 . Of 10,199 COPDGene participants with at least 10 or more pack-years of smoking and no significant interstitial lung disease or bronchiectasis, 1,260 of them had PRISm at the enrollment visit. After excluding one individual with no available post-bronchodilator spirometry, 121 with no TLC CT measures and 7 individuals with FVC > TLC CT , 1,131 participants were included in the analysis. The median value of FVC/TLC CT was 0.59 (IQR = 0.53-0.66). Of these 1,131 participants, 617 of them had acceptable spirometry measurements at the 5-year follow-up visit, 967 had available data regarding respiratory exacerbations, and 960 had vital status data available.

Results
Baseline characteristics at the enrollment visit (n = 1,131). Table 1 shows the characteristics of participants by FVC/TLC CT quartile. Age, BMI, pack-years smoking exposure, mMRC and SGRC scores, % emphysema and gas trapping, and % functional small airways disease increase with decreasing FVC/TLC CT . An increased proportion of females and decreased proportion of African Americans were associated with decreasing FVC/TLC CT . Participants in the lower FVC/TLC CT quartiles a higher prevalence of comorbidities.
Progression to COPD at 5-year follow-up. Among participants with valid spirometry at the 5-year follow up visit (n = 617), approximately 35.9% (56 of 156) of individuals in very low FVC/TLC CT quartile progressed to COPD, while 23% (37 of 160), 22% (35 of 156), and 17% (25 of 145) of individuals in the low, high, and very high FVC/TLC CT quartiles, respectively, progressed to COPD ( Fig. 2; Cochran-Armitage test for trend p < 0.001). In the multivariable-adjusted analysis, the very low FVC/TLC CT quartile at enrollment was associated with progression to COPD with an OR of 2.67 (95% CI = 1.45-5.00; p < 0.001) relative to the highest quartile (Supplementary Table S2).
Longitudinal changes in spirometry, functional capacity, and radiographic features. Supplement Table S2 shows changes in spirometry, functional capacity, and radiographic features between enrollment and follow-up visit. In the adjusted analysis, the very low FVC/TLC CT quartile at enrollment was associated with increase of 2.74% radiographic gas trapping (95% CI = 0.55-4.93; p = 0.014) relative to the highest quartile (Supplementary Table S3). FVC/TLC CT was not associated with change in FEV1, 6-MWT distance or % emphysema over time. There were no differences in the rate of decline in FEV1 by current smoking status at enrollment (combined and by FVC/TLC quartile -data not shown). , and 20% (45 of 223) of participants, respectively, had at least one severe respiratory exacerbation, with a trend towards significance (Cochran-Armitage p = 0.095). We created multivariable zero-inflated negative binomial models to examine the association of FVC/TLC CT quartile with respiratory exacerbations (Fig. 3). The very low FVC/TLC CT quartile was associated with increased relative risk for total exacerbations (IRR = 1.65; 95% CI = 1.07-2.54; p = 0.023) and severe (IRR = 2.24; 95% CI = 1.29-3.89; p = 0.004) exacerbations relative to the "very high" FVC/TLC quartile (Supplementary Table S4 Fig. 4.

Respiratory exacerbations.
In Cox proportional hazards models adjusted for age at enrollment, sex, race, BMI, current smoking at enrollment, cumulative smoking exposure, diabetes, history of asthma and congestive heart failure, increased mortality in the high quartile with a trends towards an increased mortality in the low and very low quartiles relative to the very high quartile was observed ( Table 2). In a Cox proportional hazards model examining individuals in the very high quartile relative to all other quartiles (high, low, very low) combined, a reduced risk of mortality was observed (HR = 0.53, 95% CI = 0.28-0.97,p = 0.040).  Table 1. Baseline characteristics of smokers with preserved ratio impaired spirometry across postbronchodilator forced vital capacity/total lung capacity ratio (FVC/TLC CT ) quartiles (n = 1,131). § For % GT and FRC CT % analysis, data were available for 936 subjects. ‡ For PRM data analysis, data were available for 932 subjects. # For 6-MWT data analysis, data were available for 1,121 subjects. BDR = bronchodilator response; BMI = body mass index; CAD = coronary artery disease; CHF = congestive heart failure; DM = diabetes mellitus; FRC CT % = functional residual capacity % predicted; HTN = hypertension; ICS = inhaled glucocorticosteroids, LABA = long-acting beta-agonist, LAMA = long-acting muscarinic antagonist, mMRC = modified Medical Research Council dyspnea score; OSA = obstructive sleep apnea; post-FEV1% = postbronchodilator FEV1% predicted; post-FVC% = post-bronchodilator FVC% predicted; PRM fSAD = parametric response mapping functional small airways disease; SD = standard deviation; SGRQ = St. George's Respiratory Questionnaire score; TLC CT % = total lung capacity % predicted and 6-MWD = 6-min walk test.

FVC/TLC Quartile
Scientific RepoRtS | (2020) 10:5169 | https://doi.org/10.1038/s41598-020-61932-0 www.nature.com/scientificreports www.nature.com/scientificreports/ Sensitivity analysis. When PRISm was defined using LLN criteria, we observed similar findings with those in main analysis except that FVC/TLC CT was significantly associated with chronic bronchitis, increased mMRC and SGRQ and the association with mortality was attenuated (Supplementary Tables S5-S9 and Fig. S1). . All models were adjusted for the following co-variates: age, sex, race, smoking status, smoking pack-years, body mass index (BMI), history of asthma and congestive heart failure, and diabetes mellitus. FVC/TLC CT quartile is plotted on the x-axis while the regression coefficient (and 95% CI) for each category is plotted on the y-axis. * For % GT analysis, n = 936 subjects. † For PRM fsad data analysis, n = 932 subjects. # For 6-min walk test distance analysis, n = 1,121 subjects.

Discussion
Our study explores the utility of FVC/TLC CT ratio in former and current smokers with PRISm as a potential tool to identify individuals with features of and possible increased risk for progression to obstructive lung disease. In our cohort, very low FVC/TLC CT was associated with radiographic findings traditionally associated with COPD as well as progression to COPD and respiratory exacerbations while very high FVC/TLC CT was associated with reduced mortality.
PRISm is a common spirometric pattern with a prevalence between 5% and 20% [22][23][24][25] . Although often referred to as a "restrictive spirometric pattern", 30-40% of patients with PRISm do not have reduced TLC 26,27 . On average, individuals PRISm have higher BMI, but obesity alone does not decrease vital capacity or TLC below the LLN in most individuals 28 . Notably, only about 5% of patients undergoing bariatric surgery for extreme obesity have PRISm at preoperative assessment 29 .
PRISm is comprised of a heterogeneous population with a wide range of BMI, degree of lung function impairment, and radiographic emphysema likely due to different underlying pathological processes in each individual 2 . Subgroups within PRISm may have increased risk for FEV 1 decline, progression to COPD, exacerbations, and . For exacerbation analysis, data for 967 subjects with PRISm at enrollment were available. Zero-inflated negative binomial regression models with post-bronchodilator FVC/ TLC CT as independent variable (exposure) and total exacerbations and severe exacerbations as the dependent variables (outcome) were performed. All regression models included the following co-variates: age, sex, race, body mass index, smoking status at the enrollment, smoking pack-years, history of asthma and congestive heart failure, and chronic bronchitis in the count negative binomial regression and an intercept-only model in the zero component. Follow-up time was included as an offset in the models. FVC/TLC CT quartile is plotted on the x-axis while the IRR (and 95% CI) for each category is plotted on the y-axis. IRR = incidence rate ratio, FVC/ TLC CT = forced vital capacity/total lung capacity. www.nature.com/scientificreports www.nature.com/scientificreports/ mortality. In this manuscript, we utilize FVC/TLC, which decreases in obstructive lung disease 11 , as a conceptual surrogate for RV (which was not directly measured in our cohort) to identify individuals with features obstructive lung disease within PRISm. Our finding that individuals with PRISm with low FVC/TLC have increased radiographic emphysema and gas trapping complements work from the SPIROMICS cohort, where that RV/TLC was shown to be associated with increased radiographic emphysema and gas trapping in smokers with normal lung function 30 . Apart from the fact that RV/TLC was not available in our cohort, we used FVC/TLC as it may be more sensitive to identify the presence of small airway disease than RV/TLC because FVC, a dynamic measure obtained at forced expiration, will capture dynamic collapse and air trapping not present during slow exhalation maneuver 31,32 . Future studies should examine the role of RV/TLC in PRISm. In addition, we did not examine FRC/TLC as FRC can be reduced remarkably in obesity 28 which may render difficult to interpret those measures when an obstructive lung diseases coexists.
Our findings suggest that low FVC/TLC CT may be a possible a marker of early obstructive pulmonary disease. Nevertheless, participants in the very low FVC/TLC CT quartile have higher BMI; this contrasts with the common knowledge that patients with established obstructive pulmonary disease have often lower BMI. Previous studies have shown an inverse relationship of BMI with mortality in COPD, known also as the "obesity paradox" with confounders such as exercise capacity and muscle mass possibly contributing towards favorable outcomes 33,34 . In addition, despite the fact that obesity does not typically reduce the FVC below the LLN in subjects without lung disease 29 , higher BMI decreases FVC and increases the FEV 1 /FVC ratio which can lead to underdiagnosis of obstructive pulmonary disease 31,35 . In COPD subjects with established airflow obstruction, increasing BMI is associated with higher FEV 1 /FVC 35 .
We acknowledge that the fixed threshold FEV1/FVC < 0.7 diagnostic criterion for COPD endorsed by Global Initiative for Chronic Obstructive Lung Disease(GOLD) may have also misclassified individuals with obstructive lung disease as PRISm. In a recent large population-based sample (n = 24,207), Bhatt and colleagues showed that the discriminative accuracy of FEV 1 /FVC < 0.7 to predict COPD-related death and/or hospitalization was not inferior to FEV 1 /FVC < LLN 36 . We assert that because the majority of our findings remained robust on sensitivity analyses using LLN-defined lung function categories. FVC/TLC ratio can be utilized to identify individuals with features of obstructive lung disease regardless of whether fixed-threshold or LLN criteria are used.
In COPDGene, 40.5% of PRISm individuals and 32.5% of smokers with normal lung function are African American 2 . Differences in the reliability of prediction equations may lead to the "overdiagnosis" of African American with PRISm in the absence of true pathology; this may contribute to the lower rates of African Americans in the low FVC/TLC CT quartiles. It is also unclear why females were relatively over-represented in the lower FVC/TLC CT quartiles. Whether PRISm represents a gender-specific pathway to COPD, or whether traditional FEV 1 /FVC criteria systematically misclassify women with COPD is not known 37,38 . Our sensitivity analysis using gender and race specific spirometric criteria to define PRISm showed similar findings. Future studies, especially in cohorts of diverse ancestry and ethnicity, are warranted to further explore these findings.
Previous studies have shown that air trapping is associated with FEV 1 decline. In current and former smokers with at least 20 pack-years smoking and normal lung function, RV/TLC is associated with FEV 1 decline 30 . We have extended these finding by showing that air trapping (low FVC/TLC) in individuals with PRISm is associated with progression to COPD. General population studies have also shown that individuals with abnormal non-obstructed lung function are at risk for developing COPD 39,40 . It may seem counterintuitive that low FVC/ TLC CT in PRISm was associated with progression to COPD and respiratory exacerbations, but was not associated with FEV 1 decline, increase in emphysema, and change in 6-MWT distance over time 41 . Within COPDGene, individuals with PRISm are at increased risk for respiratory exacerbations relative to smokers with normal lung function 42 . However, respiratory exacerbations in PRISm do not result in significant excess lung function decline in FEV 1 as observed in individuals with established airflow limitation 43 . A survivor bias may also be present in FEV1 decline analysis 43 . Participants that had poor lung function and low FVC/TLC may have died before the follow-up visit. Similarly, we may have not observed changes in 6-MWT distance likely due to the high variability of the test 44,45 .
Population-based studies have shown that PRISm is associated with increased cardiac 7 and all-cause mortality 1, 22 . While the increased average BMI in PRISm as a whole may mediate some of the risk associated with increased mortality, the association between very high FVC/TLC CT and lower mortality relative to all other quartiles despite concurrent adjustment for BMI, congestive heart failure, and diabetes status in our study suggests  Table 2. Associations of post-bronchodilator forced vital capacity/total lung capacity (FVC/TLC CT ) quartiles at enrollment with mortality in smokers with Preserved Ratio Impaired Spirometry (PRISm; n = 960). Cox Hazard regression models with post-bronchodilator FVC/TLCCT quartiles as independent variables (exposure) and mortality as the dependent variable(outcome) were performed. All models for mortality included the following co-variates: age, sex, race, smoking status, smoking pack-years, body mass index (BMI), history of asthma and congestive heart failure, and diabetes mellitus. HR = Hazard Ratio.